ctELM: Decoding and Manipulating Embeddings of Clinical Trials with Embedding Language Models

Brian Ondov*, Chia-Hsuan Chang*, Yujia Zhou, Mauro Giuffrè, Hua Xu · 2026

Read ctELM: Decoding and Manipulating Embeddings of Clinical Trials with Embedding Language Models by Brian Ondov*, Chia-Hsuan Chang*, Yujia Zhou, Mauro Giuffrè, Hua Xu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-18796.txt.

Description

Brian Ondov and four colleagues teach text models to read, walk, and turn back the codes that stand for trial texts in health work. Such codes crush a long trial text into a short string of counts. The crush is one-way in most tools: staff can make the code but can't read what it holds or write a new trial from it. That blind spot cuts trust and blocks fresh uses, such as to draft a trial like past ones.

The team pair a text model with the codes of health trials. The pair learns to turn a code back to words and to bend a code by a text ask. A two-way code could let staff search trials by feel, not just by key words. It could also let them draft new trials from the form of old ones.

Catalog record

CREATORBrian Ondov*, Chia-Hsuan Chang*, Yujia Zhou, Mauro Giuffrè, Hua Xu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE72 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-18796

Browse the full library · About the Tower of Babel